RESEARCH ARTICLE Open Access

Prognostic potential of pre-partum bloodbiochemical and immune variables forpostpartum mastitis risk in dairy cowsRuo-wei Guan, Di-ming Wang, Bei-bei Wang, Lu-yi Jiang and Jian-xin Liu*

Abstract

Background: Mastitis is the most frequent diseases for transition cows. Identification of potential biomarkers fordiagnosis of mastitis is important for its prevention. Thus, this study was conducted to investigate blood variablesrelated to lipid metabolism, oxidative stress and inflammation, and serum variables that are related to health inpostpartum cows.

Results: Seventy-six healthy Holstein dairy cows at week 4 before calving were selected to collect blood samplesfrom weeks − 4 to 4 weekly relative to calving, respectively. Milk yield and composition were recorded weekly.According to the cut-off of somatic cell counts (SCC) for diagnosis of mastitis, 33 cows with SCC ≥ 500,000 cellsml− 1, 20 cows with 200,000 cells ≤ SCC < 500,000 cells ml− 1, and 23 cows with SCC < 200,000 cells ml− 1 weredefined as high, middle, and low SCC, respectively. Serum concentrations of β-hydroxybutyrate were higher (P <0.01) during all weeks, and non-esterified fatty acids were higher in high SCC than in low SCC cows from weeks − 3to 2 relative to calving. Higher serum concentrations of superoxide dismutase (P < 0.01) and lower malondialdehydelevels (P < 0.01) in low SCC than in high SCC cows indicate that the latter suffered from oxidative stress. Thedifference analysis of the three groups suggested that none of the above-mentioned variables can be used aspotential prognostic candidates. On the other hand, high SCC cows exhibited higher blood neutrophil tolymphocyte ratio (NLR, P < 0.01) and platelet to lymphocyte ratio (PLR, P < 0.01) than low SCC cows, with a higherNLR (P < 0.01) in middle SCC than in low SCC cows. The high SCC cows had lower levels of anti-inflammatoryfactors including IL-10 (P = 0.05), but higher levels of proinflammatory factors such as IL-6 (P < 0.01), TNF-α (P < 0.05),and PSGL-1 (P < 0.01) than low SCC cows.

Conclusions: The significantly different NLR and PLR pre-partum between the middle and low SCC cows suggesttheir prognostic potential for postpartum mastitis risk.

Keywords: Blood biochemical variables, Neutrophil to lymphocyte ratio, Platelet to lymphocyte ratio, Mastitis,Transition cows

© The Author(s). 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License,which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you giveappropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate ifchanges were made. The images or other third party material in this article are included in the article's Creative Commonslicence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commonslicence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtainpermission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to thedata made available in this article, unless otherwise stated in a credit line to the data.

* Correspondence: liujx@zju.edu.cnInstitute of Dairy Science, College of Animal Sciences, Zhejiang University,Hangzhou 310058, P. R. China

Guan et al. BMC Veterinary Research (2020) 16:136 https://doi.org/10.1186/s12917-020-02314-6

BackgroundThe periparturient period is a critical stage for lactatingdairy cows. During this stage, cows suffer from a seriesof physiological and metabolic disorders, including re-duced feed intake, and enhanced fat mobilization, re-flective of changed lipid metabolism analytes such asnon-esterified fatty acids (NEFA) and β-hydroxybutyricacid (BHB), leading to enhanced production of reactiveoxygen species (ROS) [1]. Many previous studies agreethat significant changes in oxidative stress are observedduring the transition period [1, 2], with activity of super-oxide dismutase (SOD), glutathione peroxidase (GSH-Px), and malondialdehyde (MDA) as the representativeanalytes for oxidative stress. The impact of oxidativestress during the transition period may be a majorunderlying factor of inflammatory and immune dysfunc-tion in dairy cattle, further resulting in metabolic dis-eases such as mastitis during the post-partum stage [3].Mastitis is a common disease in non-ruminant and ru-

minant animals and can be triggered by milk accumula-tion in mammary gland (MG), and invasion of bacteriasuch as Staphylococcus aureus, finally leading to inflam-mation [4]. The milk somatic cell counts (SCC) is a sen-sitive biomarker of mammary inflammation [5, 6]. TheInternational Dairy Federation (IDF) has recommendedto establish the thresholds as the cut-off of SCC for diag-nosis of mastitis [7, 8]. Variations in SCC depend mainlyon the recruitment of leukocytes from blood to MG andfinally to milk, most often in response to an inflamma-tory reaction in the mammary tissue caused by thepathogenic bacterial intrusion [9]. The high incidence ofmastitis has always been a problem that needs to besolved. Early monitoring for mastitis is an effective wayto prevent its development, but information is limited inthis area. Previous studies found evidence of a slightlyincreased overall breast cancer risk in women with a his-tory of mastitis [10]. Cancer-related inflammatory cellscan secrete a series of inflammatory mediators, such ascytokines, and promote tumor proliferation, invasion,and metastasis. This inflammatory response can bereflected by some blood indicators, such as bloodneutrophil-to-lymphocyte ratio (NLR), and platelet-to-lymphocyte ratio (PLR) [11]. Therefore, both NLR andPLR have been recognized as effective indicators of sys-temic inflammatory response and have good prognosticvalue for some malignant diseases [12].The immune system in the MG is a defense mechan-

ism involving innate and acquired immunity initiated bythe MG to resist bacterial invasion [13]. The papillarytube is the physical and first line defense against bacteria[14]. When the bacteria enter into the MG, white bloodcells (WBC) such as polymorphonuclear neutrophils(PMN) are activated through cytokines such as interleu-kin (IL)-6 and tumor necrosis factor (TNF)-α [15], and

are aggregated by chemical induction, causing an oxida-tive burst and phagocytosis of the pathogen [16]. Mean-while, IL-6, TNF-α, and IL-10 can affect the function ofPMN by regulating their apoptosis [17]. Recent studieshave shown that PMN require P-selectin glycoproteinligand-1 (PSGL-1) to activate them on the vascularendothelium. The PMN interact with PSGL-1 on plate-lets to migrate in blood vessels and exert anti-inflammatory effects [18]. When the pathogen is notkilled by the innate immune defense, acquired immunitywill be activated to secrete CD4 and CD8, whichrecognize and inactivate antigen molecules [19].Metabolic disorders decline the body’s resistance,

while pathogen invasion activates the MG immune sys-tem. Mammary immunity affects the function and quan-tity of immune cells, resulting in differences in NLR andPLR. We hypothesized that cows with high SCC willhave alterations in their blood and serum variables be-fore calving, increasing the risk of postpartum mastitis.The objective of the current study was to investigateblood and serum biochemical variables in the transitioncows, and to evaluate the prognostic potential of prepar-tum variables for postpartum mastitis risk.

ResultsMilk yield and compositionThe SCC average from 4 weeks for each cow was usedto classify each cow into different SCC groups. Accord-ing to the cut-off of SCC for diagnosis of mastitis, 33cows with SCC ≥ 500,000 cells ml− 1, 20 cows with 200,000 cells ≤ SCC < 500,000 cells ml− 1, and 23 cows withSCC < 200,000 cells ml− 1 were defined as high (HSCC),middle (MSCC), and low SCC (LSCC), respectively.These classifications were not based on mastitic signs,but are reflective of mastitis risk.Milk yield was lower (P < 0.01), but milk protein con-

tent was higher (P = 0.02) in cows with HSCC than inLSCC cows (Table 1). Content of milk urea nitrogen waslower in the HSCC cows than in LSCC cows (P < 0.01).The average SCC was 1460.5, 346.0, and 77.6 × 103 cellsml− 1 in the HSCC, MSCC, and LSCC cows, respectively.

Lipid metabolism and oxidative stress AnalytesDuring the whole experimental period, HSCC cows hadhigher (P < 0.01, Fig. 1a) serum concentrations of β-hydroxybutyric acid (BHBA) than LSCC cows, with aninteraction effect (P < 0.01) of health status and week.Serum concentration of non-esterified fatty acid (NEFA)in HSCC cows was greater (P < 0.01, Fig. 1b) than inLSCC cows at all the sampling weeks, except week 4.The MSCC cows had higher (P < 0.01) BHBA concentra-tions than the LSCC cows did at weeks − 4, − 3 andweek 1.

Guan et al. BMC Veterinary Research (2020) 16:136 Page 2 of 11

Cows with HSCC had lower activity of SOD (P < 0.01,Fig. 1c) and GSH-Px (P < 0.01, Fig. 1d) compared withLSCC cows during the whole period, except week − 2.The HSCC cows had greater ROS (P < 0.01, Fig. 1e) atweeks 1 and 2, and greater activity of MDA (P < 0.01,Fig. 1f) during the whole period, than LSCC cows. Thesignificant effects of interaction between health statusand week were found (P < 0.01, Fig. 1) on serum SOD,GSH-Px, ROS, and MDA.

Platelets and WBC quantificationNumbers of lymphocytes was lower (P < 0.01, Fig. 2c) incows with HSCC than in LSCC cows with lower num-ber of lymphocytes (P < 0.01) at weeks − 4 and − 3 inMSCC cow than in LSCC cows, while the PMN, NLRand PLR were greater (P < 0.01, Fig. 2a, e, f) in HSCCcows than in LSCC cows during the whole period. Thenumber of platelets was lower (P < 0.01, Fig. 2d) inHSCC than in LSCC group until week 2. The cows withHSCC had a lower number of total WBC throughoutthe study (P = 0.01, Fig. 2b) than LSCC cows. The NLRhad an effect of interaction between health status andweek (P = 0.03, Fig. 2e). The MSCC cows had lowerNLR (P < 0.05) at week 4 than the LSCC cows. At week− 4, cows with MSCC had greater PLR than LSCC cows(P < 0.01, Fig. 2f).

Serum inflammatory cytokines and CD4 and CD8The content of IL-6 was higher (P < 0.01, Fig. 3c) in theHSCC cows during the whole period, with higher (P <0.05) content of IL-6 in the MSCC cows than in LSCCcows at week − 4, − 3, 1, 1 and week 4. The HSCC cowsalso had higher (P < 0.05, Fig. 3d) TNF-α at week − 4, − 3,− 1, 1, and 4 than the MSCC and LSCC cows, but no dif-ference (P > 0.05) observed between the two groups. The

content of IL-10 was lower (P < 0.05, Fig. 3e) in HSCCand MSCC cows than in LSCC cows at week − 3, − 1, 1, 2,and 4; and lower (P = 0.03) in HSCC cows than in MSCCcows at week 2 and 4. Serum PSGL-1 content was higher(P < 0.01, Fig. 3f) in HSCC and MSCC cows than in LSCCcows during the whole experimental period, with nodifference (P > 0.05) between the HSCC and MSCC cowsexcept week 4.Concentration of CD8 was lower in HSCC cows than

in LSCC cows (P < 0.01, Fig. 3a), with lower values inMSCC cows than in LSCC cows at week 1 (P = 0.03) and2 (P = 0.01) postpartum. Serum CD4 concentration waslower in HSCC cows than in LSCC cows during thewhole period except week 4 (P < 0.01, Fig. 3b), withhigher value at week − 4 in MSCC cows than in HSCCcows (P < 0.01).

Correlation analysis and ROC curvesCorrelations between serum BHBA or NEFA levels inprepartum and milk SCC are shown in Fig. 4. Overall,milk SCC were positively correlated with pre-partumserum BHBA (r = 0.51, P < 0.01, Fig. 4a) and serumNEFA (r = 0.45, P < 0.01, Fig. 4b), respectively. SerumBHBA concentration was positively correlated to SCC,with coefficients of 0.65 (week − 4, Figure S1a), 0.73(week − 3, Figure S1b) and 0.72 (week − 1, Figure S1c).The coefficients of correlation between serum NEFAand SCC in cows were 0.75 (week − 4, Figure S2a), 0.72(week − 3, Figure S2b), and 0.75 (week − 1, Figure S2c).Overall, milk SCC were positively associated with the

NLR (r = 0.73, P < 0.01, Fig. 5a) and PLR (r = 0.74, P <0.01, Fig. 5b) before calving. Milk SCC had positivecorrelations with NLR (P < 0.01) at week − 4 (r = 0.81,Figure S3a), − 3 (0.79, Figure S3b), and − 1 (r = 0.74,Figure S3c), and with the PLR (P < 0.01) at week − 4 (r =

Table 1 Basic information and lactation performance of dairy cows with low, middle, and high somatic cell counts (SCC)

Items Groups1 SEM P-value2

LSCC MSCC HSCC Hs Wk Hs*Wk

Number of cows (head) 23 20 33 – – – –

Parity 2.8 2.9 2.7 0.17 0.26 – –

SCC (103/mL) 77.6 346.0 1460.5 – – – –

Milk yield (kg) 34.3a 31.4b 26.5b 0.71 < 0.01 < 0.01 0.96

Milk composition (%)

Milk fat 4.85 5.07 4.88 0.11 0.36 < 0.01 0.53

Milk protein 3.24b 3.26b 3.35a 0.03 0.02 < 0.01 0.48

Milk lactose 4.96b 4.97a 4.80b 0.03 < 0.01 < 0.01 0.75

Total solids 13.2 13.4 13.2 0.12 0.36 < 0.01 0.72

Milk urea nitrogen (mgN/dL) 12.2a 10.4b 8.5b 0.52 < 0.01 0.43 0.951LSCC = cows with SCC (SCC < 200,000 cells ml− 1, n = 23); MSCC = cows with middle SCC (200,000 cells ≤ SCC < 500,000 cells ml− 1, n = 20); HSCC = cows with highSCC (SCC ≥ 500,000 cells ml− 1, n = 33)2Effect of health status (Hs), sampling week (Wk), and interaction of health status by sampling week (Hs ×Wk)a,bMeans within the same row followed by different superscripts differ at P < 0.05

Guan et al. BMC Veterinary Research (2020) 16:136 Page 3 of 11

0.77, Figure S4a), − 3 (r = 0.74, Figure S4b), and − 1 (r =0.72, Figure S4c).The receiver-operator characteristic (ROC) curve of

NLR and PLR showed that the area under the curve(AUC) of the NLR in HSCC cows were 0.880 (week − 4,Figure S5a), 0.853 (week − 3, Figure S5b), and 0.883(week − 1, Figure S5c), and the AUC of the PLR were0.848 (week − 4, Figure S5a), 0.733 (week − 3, FigureS5b), and 0.765 (week − 1, Figure S5c). The AUC of theNLR in the MSCC cows were 0.765 (week − 4, FigureS6a), 0.625 (week − 3, Figure S6b), and 0.740 (week − 1,Figure S6c), and the AUC of PLR were 0.850 (week − 4,Figure S6a), 0.583 (week − 3, Figure S6b), and 0.635(week − 1, Figure S6c). In both HSCC and MSCC cows,the AUC of the NLR was higher than that of the PLR.

The ROC curve analysis results for BHBA, NEFA (FigureS7, S8), SOD, GSH-Px, ROS, MDA (Figure S9, S10) inHSCC cows and LSCC cows and in MSCC and LSCCcows are shown in the supplementary figures.

DiscussionPeriparturient dairy cows usually suffer from metabolicstress and had increasing serum concentrations of BHBAand NEFA [20]. Cows with higher SCC had higherBHBA and NEFA levels in serum than LSCC-animals inthe current study, with the rapidly increased prepartumBHBA levels with time, indicating that perinatal ketosismay have some association with the development ofmastitis, consistent with a pervious study in which hep-atic dysfunction in pre-partum dairy cows could be

Fig. 1 Changes of lipid metabolism and oxidative stress analytes in dairy cows with different somatic cell counts (SCC). -Δ-, high SCC (SCC≥500,000 cells ml− 1, n = 33); -■-, middle SCC (200,000 cells ≤ SCC < 500,000 cells ml− 1, n = 20); -○-, low SCC (SCC < 200,000 cells ml− 1, n = 23).BHBA, β-hydroxybutyrate; GSH-Px, glutathione peroxidase; MDA, malondialdehyde; NEFA, non-esterified fatty acid; ROS, reactive oxygen species;SOD, superoxide dismutase. Hs = effect of health status; Wk = effect of sampling week; Hs * Wk = interaction effect (Hs ×Wk). * Hs (P < 0.05)among three groups; **Hs (P < 0.01) among three groups. Error bars indicate SEM

Guan et al. BMC Veterinary Research (2020) 16:136 Page 4 of 11

associated with post-partum mastitis [21]. Metabolicdisorders in dairy cows cause increasing oxidative energyconsumption in liver and MG [22]. From our results,HSCC cows had lower activity of SOD and GSH-Px thanLSCC cows, indicating greater oxidative stress in HSCCcows [21, 23]. The results of ROC analysis of BHBA andNEFA showed that both had association with the risk inclinical or subclinical mastitis. However, the elevated BHBAand NEFA levels are mainly related to ketosis in dairy cowsand were not reflective of perinatal cows experiencingmastitis [24]. Such levels exert relatively minimal alterationsto immune function [24] or cannot directly reflect theprocess of inflammation [25]. Therefore, we did not chooseplasma variables (NEFA, BHBA, SOD or GSH-Px) as prog-nostic candidates for post-partum mastitis.

Severe oxidative stress and high mobilization of bodyreserves in dairy cows may lead to reduced immunefunction and increase the risk of postpartum mastitis[25]. It is well known that WBC (PMN, lymphocytes,etc.) and PLT are important immune cells in the blood[26]. However, the difference analysis of the three groupssuggested that none of the above-mentioned variablescan be used as potential prognostic candidates under thecurrent experimental conditions. The derangement ofthe WBC and PLT are known as poor independentprognostic factors for many solid tumors [27]. On theother hand, blood NLR has been used as one of the indi-cators to evaluate the body’s immune system, a strongprognostic factor of certain cancers [13], and as a pre-dictor of inflammatory or infectious lesions [28] and

Fig. 2 Changes of platelets and white blood cells relative to health in dairy cows with different somatic cell counts (SCC). -Δ-, high SCC (SCC≥500,000 cells ml− 1, n = 33); -■-, middle SCC (200,000 cells ≤ SCC < 500,000 cells ml− 1, n = 20); -○-, low SCC (SCC < 200,000 cells ml− 1, n = 23).Lymph, lymphocytes; PMN, polymorphonuclear neutrophils; PLT, platelets; WBC, white blood cells; NLR, neutrophil-to-lymphocyte ratio; PLR,platelet-to-lymphocyte ratio. Hs = effect of health status; Wk = effect of sampling week; Hs * Wk = interaction effect (Hs ×Wk). * Hs (P < 0.05)among three groups; **Hs (P < 0.01) among three groups. Error bars indicate SEM

Guan et al. BMC Veterinary Research (2020) 16:136 Page 5 of 11

Fig. 3 Changes of Serum inflammatory Cytokines and CD4 and CD8 in dairy cows with different somatic cell counts (SCC). , high SCC (SCC≥ 500,000cells ml− 1, n=33); , middle SCC (200,000 cells ≤ SCC< 500,000 cells ml− 1, n=20); , low SCC (SCC< 200,000 cells ml− 1, n= 23). IL, interleukin; TNF,tumor necrosis factor; PSGL-1, p-selectin glycoprotein ligand-1. Error bars indicate SEM. a, b, c: Different letters within the same week differ (P<0.05)

Fig. 4 Correlation between somatic cell count (SCC) and prepartum serum concentrations of β-hydroxybutyric acid (BHBA, a) or non-esterifiedacid (NEFA, b) in dairy cows (n = 76)

Guan et al. BMC Veterinary Research (2020) 16:136 Page 6 of 11

postoperative complications [29]. In addition, blood PLRis a novel inflammatory marker, and has been demon-strated to be a predictor of various cardiovascular dis-eases and tumors [30], and the risk for critical limbischemia [31]. Correlation analysis suggested that bothprepartum NLR and PLR were positively associated withmilk SCC in dairy cows in the current study, suggestingan increased risk of postpartum mastitis in cows withhigher NLR and PLR before calving. One possibility isthat neither PMN nor WBC alone can be a prognosticindex, as there exists a balance between lymphocytesand neutrophils, and a balanced status of inflammatoryresponse [32]. Similarly, a balance exists between plateletand lymphocyte counts in blood. An increased plateletcount will accelerate bacterial migration, while a de-creased lymphocyte count will reduce the animals’ abilityto resist inflammation. A combination of these two factorsmay be more comprehensive in the status of inflammationincidence [18]. In this study, the relatively lower concen-tration of CD4 and CD8 in the serum of animals withHSCC indicated their dysfunction in such dairy cows,which further lowered the ability of lymphocytes to resistdiseases and increased the risk of postpartum mastitis indairy cows. This is consistent with the results by Koyaet al. [33], who reported reduced CD4 and CD8 T cells inperipheral blood of patients with non-lactation mastitis.Changes in blood cytokine levels are predominant reg-

ulators of blood NLR and PLR alteration in pre-partumdairy cows [13–15]. Both IL-6 and TNF-α are importantpro-inflammatory factors that affect PMN formation andrelease [34]. Cytokine IL-6 inhibits apoptosis of PMN inpatients with osteomyelitis [35] and could inhibit thenatural apoptosis of endogenous neutrophils [36]. Onthe other hand, TNF-α promotes the apoptosis of

peripheral blood neutrophils in rats [37]. In the presentstudy, the level of IL-6 was significantly different amongthe three groups, and may inhibit the apoptosis of PMNand thus increase the number of PMN. In general, ele-vated levels of IL-6 in the serum of pre-partum HSCCcows, compared to those of LSCC cows, would increasethe release and activation of PMN, inhibit the apoptosisof PMN, then increase the number of PMN and NLR.The IL-10 can inhibit the synthesis of pro-inflammatorycytokines by T cells and monocytes/macrophages [38].In the present study, HSCC cows had lower levels ofpre-partum serum IL-10 than LSCC cows, leading todecreased inflammation resistance, and an increasedpostpartum mastitis risk. In addition, IL-10 can promotethe apoptosis of PMN [39]. Compared with the LSCCcows, HSCC cows had lower IL-10 activity in blood, andpossibly exhibited reduced PMN apoptosis, and in-creased PMN and NLR levels in blood. It is known thatneutrophils interact with PSGL-1 on platelets to migratein blood vessels and exert anti-inflammatory effects;PSGL-1 knock out mice had reduced numbers of bloodneutrophils migrating along the inner surface of bloodvessel [18]. The higher levels of PSGL-1 in the HSCCcows in this study suggested that PMN of HSCC cowsmay be activated to increase adhesion, which further ele-vates the expression of TNF-α [26]. In summary, exces-sive lipid mobilization of cows caused higher serumconcentration of NEFA and BHBA, and lower SOD, andGSH-Px but higher MDA, resulting in the increased oxi-dative stress. Excessive oxidative stress would destroythe cow’s immune system. Increased NLR and PLR inHSCC cows indicate that the cows were in an inflamma-tory status and could be a result of lower IL-10, andhigher IL-6, TNF-α, and PSGL-1 (Fig. 6).

Fig. 5 Correlation between somatic cell count (SCC) and prepartum neutrophil-to-lymphocyte ratio (NLR, a) or platelet-to-lymphocyte ratio (PLR,b) in dairy cows (n = 76)

Guan et al. BMC Veterinary Research (2020) 16:136 Page 7 of 11

From the ROC curve analysis, it is indicated that bothper-partum NLR and PLR had good predictive potentialin higher SCC in post-partum dairy cows, indicating anincreased risk of mammary disease. A higher AUC ofthe NLR than the PLR suggested that NLR may be an in-dicator better than the PLR,According to the prevalence of the disease in the farm

evaluated, the sample size is limited in the current studyas a preliminary work. In a subsequent study, the samplesize could be expanded to validate the results of thisstudy.

ConclusionPost-partum cows with HSCC and MSCC had greaterbody fat mobilization and increased oxidative stress dur-ing their pre-partum stage, compared with LSCC cows.According to the ROC curve analysis, NLR and PLR hadprognostic potential for postpartum risk in clinical andsubclinical mastitis. Our study suggested that pre-partum monitoring blood variables such as NLR andPLR can be an efficient method to predict the incidence

of clinical and sub-clinical mastitis risk in post-partumdairy cows, helping dairy producers to treat such cowsin advance to prevent mastitis risk.

MethodsAnimals and managementThe experimental procedures used in this study were ap-proved by the Animal Care Committee of Zhejiang Uni-versity (Hangzhou, China) and were in accordance withthe university’s guidelines for animal research. Eighty-sixhealthy pregnant Holstein dairy cows at 4 weeks beforecalving were selected at the Hang-Jiang Dairy Farm,Zhejiang Province, China; where a total of 1500 dairycattle are reared with around 800 lactating cows. Permis-sion to collect sample was obtained from the farmowner. All the selected cows were similar in parity (2.93,SD 0.89) and body condition score (3.29, SD 0.37). Ex-periment lasted from mid-September to late November,2017. A veterinarian checked the selected cows beforethe trial and no disease record was observed. Then, cowswere monitored weekly during the whole experiment

Fig. 6 The immune processes in the dairy cows with high somatic cell counts (SCC). Excessive body fat mobilization increases the serumconcentrations of β-hydroxybutyric acid (BHBA) and non-esterified acid (NEFA), reduces the activities of serum superoxide dismutase (SOD) andglutathione peroxidase (GSH-Px), and increases the activity of serum malondialdehyde (MDA), leading to oxidative stress and changed immunestatus of the HSCC cows. The tissue innate immune systems will then be activated, with polymorphonuclear neutrophils (PMN) playing animportant role. When the pathogen is not killed by the innate immune defenses, the acquired immune system is activated, with lymphocytesplaying an important role in this process. During the whole period, the cytokines such as interleukin (IL)-6, tumor necrosis factor (TNF)-α, IL-10,and p-selectin glycoprotein ligand-1 (PSGL-1) play different roles, and increase blood neutrophil-to-lymphocyte ratio (NLR) and platelet-to-lymphocyte ratio (PLR). (+) indicates promotion, (−) indicates inhibition; I, innate immunity, II, acquired immunity; a polymorphonuclearneutrophils; b lymphocyte; c Platelet

Guan et al. BMC Veterinary Research (2020) 16:136 Page 8 of 11

period to evaluate their health status based on the clin-ical signs. As a result, ten cows were removed due totheir health problems such as clinical mastitis, milkfever, and ketosis. Thus, 76 cows were finally enrolled inthis study. Cows were housed in individual tie stalls,bedded with sawdust, milked three times daily, and hadfree access to water throughout the experiment. Thecows were kept as free stall in the barn while the experi-ment was conducted. Therefore, the animals were re-leased after the study was completed.Diets were offered as total mixed rations ad libitum

three times daily at 0630, 1300, 1830 h to allow approxi-mately 5% orts. All rations were formulated to meet thenutrient requirements of dry and early lactating cows[40]. Ingredient and chemical composition of the dietsfed during the prepartum and postpartum periods areshown in Supplementary Table S1.

Experimental design and samplingBefore the data were collected, three thresholds of SCCwere designed according to the IDF recommendations[7, 8]. Blood samples were collected during the wholeperiod from weeks − 4 to 4 relative to calving. Fromcalving to week 4, milk yield was recorded and milksamples were collected for analysis of milk compositionincluding SCC. After all the samples were collected, 76dairy cows were divided into three groups based on thepre-designed thresholds of the SCC to compare their dif-ferences in blood variables and to analyze the correlationbetween pre-partum blood or serum variables and SCC.Blood samples were obtained from the coccygeal vein

at around 0930 h – 1000 h once a week at weeks − 4, −3, − 2, − 1, 1, 2, 3, and 4 relative to calving. The bloodwas collected into 2 mL EDTA anticoagulant vacutainertubes (EDTA k2), and then sent to an animal hospital(Nargis Animal Hospital, Hangzhou, China) within 2 hfor analysis of blood variables. Another blood samplewas collected into 5 mL vacutainer tubes (Yuli MedicalInstrument Co., Jiangsu, China) and then kept at 4 °Cuntil separation of serum. Clotted blood was centrifugedat 3000×g for 15 min. The serum was transferred to asterile 1.5 mL RNase-free plastic test tube (MCT-150-C-S, Axygen, Corning, NY, USA), and then stored at −80 °C until analysis.Milk yield was recorded daily during the sampling

periods, and milk samples were collected once a week atweeks 1, 2, 3, and 4 with milk-sampling devices(Waikato Milking Systems NZ Ltd., Hamilton, NewZealand). Daily samples were composited proportionally(4:3:3 according to three times of milking) and storedwith added bronopol tablets (milk preservative; D & FControl System Inc., San Ramon, CA, USA) for lateranalysis of milk compositions.

Analysis of blood, serum and milk samplesNumber of PMN, lymphocytes, and platelet were quanti-fied using an automatic blood cell analyzer (B2600,Mandray, Shexhen, China). The NLR and PLR were thencalculated. An Auto Analyzer 7020 instrument (HitachiHigh-technologies Corporation, Tokyo, Japan) was usedto determine serum lipid metabolism analytes includingNEFA, and BHBA as well as oxidative stress analytessuch as ROS, SOD, GSH-Px and MDA, with colorimet-ric commercial kits (Ningbo Medical System Biotechnol-ogy Co., Ltd., Ningbo, China). The serum concentrationsof CD4 (Catalog No. MB-6980A), CD8 (MB-4853A), IL-6 (MB-4905A), IL-10 (MB-4931A), TNF-α (MB-4838A),and PSGL-1 (MB-9620A) were analyzed using commer-cially available ELISA kits specific for the bovine species(Jiangsu Enzyme Biotechnology Co., Ltd., Jiangsu,China). Before the samples were analyzed, an engineerfrom the instrument was requested to calibrate the in-strument with a standard to validate the analyticalprocedures.Milk samples were analyzed for milk protein, fat,

lactose, total solids, urea nitrogen and SCC using aCombi Foss FT+ instrument (Foss Electric, Hillerød,Denmark).

Statistical analysesIn order to identify the prognostic factors of clinical andsubclinical mastitis in cows, data were compared forLSCC, MSCC and HSCC cows separately at each timepoint. Data on lactation performance, blood variablesrelated to lipid metabolism, oxidative stress and inflam-mation, and serum variables were analyzed using thePROC MIXED procedure with repeated measurement(SAS Institute Inc., Cary, NC, USA). The model includedthe health status (LSC, MSCC, HSCC), sampling weeksand their interaction (health status × wk) as the fixed ef-fects, and cow within the block as a random effect.Spearman correlation analysis was conducted betweenpre-partum blood or serum variables and SCC in threegroups using GraphPad Prism 6 software (GraphPadSoftware Inc., La Jolla, CA, USA). Significance was de-clared at P < 0.05, and tendency was defined at 0.05 <P < 0.10. Graphs were generated with GraphPad Prism 6software (GraphPad).Biomarker profiles and quality of the biomarker sets

were assessed using ROC curves, as calculated by IBMSPSS Statistics 20.0 software (International Business Ma-chines Corp., Armonk, New York 10,504, USA). TheROC curves are often summarized into a single metricknown as the AUC, which is indicative of the accuracyof a test for correctly distinguishing one group fromothers. The NLR and PLR were evaluated with ROCanalysis, and the higher AUC values associated withNLR and PLR were used to determine the most

Guan et al. BMC Veterinary Research (2020) 16:136 Page 9 of 11

predictive critical threshold for disease identification.Sensitivity was defined as the proportion of animalsdiagnosed with disease that were above the threshold ofa given blood variable, and specificity was the proportionof animals that did not have disease below a giventhreshold [41]. If all positive samples are ranked beforenegative ones, the AUC is 1.0, which indicates aperfectly discriminating test.

Supplementary informationSupplementary information accompanies this paper at https://doi.org/10.1186/s12917-020-02314-6.

Additional file 1 Table S1. Ingredient and chemical composition of thediets fed during the prepartum and postpartum periods. Figure S1.Correlation between somatic cell count (SCC) and β-hydroxybutyric acid(BHBA) in dairy cows. Figure S2. Correlation between somatic cell count(SCC) and non-esterified acid (NEFA) in dairy cows. Figure S3. Correl-ation between somatic cell count (SCC) and neutrophil-to-lymphocyte ra-tio (NLR) in dairy cows. Figure S4. Correlation between somatic cellcount (SCC) and platelet-to-lymphocyte ratio (PLR) in dairy cows. FigureS5. Receiver-operator characteristic curves of blood neutrophil-to-lymphocyte ratio (NLR) and blood platelet to-lymphocyte ratio (PLR) indairy cows with low (LSCC) and high somatic cell count (HSCC). FigureS6. Receiver-operator characteristic curves of blood neutrophil-to-lymphocyte ratio (NLR) and blood platelet to-lymphocyte ratio (PLR) indairy cows with low (LSCC) and middle somatic cell count (MSCC). Fig-ure S7. Receiver-operator characteristic curves of lipid metabolism ana-lytes in dairy cows with low (LSCC) and high somatic cell count (HSCC).Figure S8. R Receiver-operator characteristic curves of lipid metabolismanalytes in dairy cows with low (LSCC) and middle somatic cell count(MSCC). Figure S9. Receiver-operator characteristic curves of serum anti-oxidative analytes in dairy cows with low (LSCC) and high somatic cellcount (HSCC). Figure S10. Receiver-operator characteristic curves ofserum anti-oxidative analytes in dairy cows with low (LSCC) and middlesomatic cell count (MSCC).

AbbreviationsAUC: Area under the curve; BHBA: β-hydroxybutyrate; GSH-Px: Glutathioneperoxidase; IL: Interleukin; MDA: Malondialdehyde; MG: Mammary gland;NEFA: Non-esterified fatty acid; NLR: Neutrophil to lymphocyte ratio;PLR: Platelet to lymphocyte ratio; PLT: Platelet; PMN: Poly-morphonuclearneutrophils; PSGL-1: P-selectin glycoprotein ligand-1; ROC: Receiver-operatorcharacteristic; ROS: Reactive oxygen species; SCC: Somatic cell counts;SOD: Superoxide dismutase; TNF: Tumor necrosis factor; WBC: White bloodcell

AcknowledgmentsThe authors acknowledge the members of the Institute of Dairy Science atZhejiang University (Hangzhou, China) for their assistance with the samplingand analysis of the samples.

Authors’ contributionsRWG, DMW, and JXL designed the trial. RWG collected, processed andanalyzed the samples, interpreted the data and wrote the manuscript; DMWplanned manuscript ideas, interpreted the data, and wrote the manuscript;BBW participated in sample collection and serum preparation; LYJ wrote themanuscript; JXL conceived and planned the research, explained the data,and revised the manuscript. All authors have read and approved the finalmanuscript.

FundingThis study was financially supported by the grants from China-USA Intergov-ernmental Collaborative Project in S & T Innovation under the National Key R& D Program (No. 2018YFE0111700), the China Agriculture Research System(No. CARS-36), and the National Nature Science Foundation of China (No.31872380). The Foundation reviewed and approved the design of the study,

but had no role in the collection, analysis, or interpretation of data or in writ-ing the manuscript.

Availability of data and materialsThe datasets used and/or analysed during the current study are availablefrom the corresponding author on reasonable request.

Ethics approval and consent to participateThe experimental procedures used in this study were approved (No.ZJU20170422) by the Animal Care Committee of Zhejiang University(Hangzhou, China), and were in accordance with the University’s guidelinesfor animal research. Permission to collect sample was obtained from thefarm owner by verbal. As the farm is our collaborate farm, the ethicscommittee approved this procedure.

Consent for publicationNot applicable.

Competing interestsThe authors declare that they have no competing interests.

Received: 22 October 2019 Accepted: 12 March 2020

References1. Colakoglu HE, Yazlik MO, Kaya U, Colakoglu EC, Kurt S, Oz B, Bayramoglu R,

Vural MR, Kuplul S. MDA and GSH-Px activity in transition dairy cows underseasonal variations and their relationship with reproductive performance. JVet Res. 2017;61:497–502.

2. Turk R, Juretić D, Gereš D, Svetina A, Turk N, Flegar-Meštrić Z. Influence ofoxidative stress and metabolic adaptation on PON1 activity and MDA levelin transition dairy cows. Anim Reprod Sci. 2008;108:98–106.

3. Barkema HW, Green MJ, Bradley AJ. The role of contagious disease in udderhealth. J Dairy Sci. 2009;92:4717–29.

4. Pereira H, Pinder SE, Sibbering DM. Pathological prognostic factors in breastcancer. IV: should you be a typer or a grader? A comparative study of twohistological prognostic features in operable breast carcinoma. Histopathol.1995;27:219–26.

5. de Haasa Y, Barkemab HW, Schukkenc YH, Veerkamp RF. Associationsbetween somatic cell count patterns and the incidence of clinical mastitis.Prev Vet Med. 2005;67:55–68.

6. Satu P. Indicators of inflammation in the diagnosis of mastitis. Vet Res. 2003;34:565–78.

7. Hillerton JE. Redefining mastitis based on somatic cell count. IDF Bulletin.1999;345:4–6.

8. IDF (International Dairy Federation). A monograph of bovine mastitis. 1971.9. Rainard P, Foucras G, Boichard D, Rupp R. Invited review: low milk somatic

cell count and susceptibility to mastitis. J Dairy Sci. 2018;101(8):6703–14.10. Lambe M, Johansson ALV, Altman D. Mastitis and the risk of breast cancer.

Epidemiol. 2009;20:747–51.11. Kwon HC, Kim SH, Oh SY. Clinical significance of preoperative neutrophil-

lymphocyte versus platelet-lymphocyte ratio in patients with operablecolorectal cancer. Biomarkers. 2012;17:216–22.

12. Forget P, Khalifa C. What is the normal value of the neutrophil-to-lymphocyte ratio. BMC Res Notes. 2017;10:12.

13. Werner-Misof C, Pfaffl MW, Meyer HHD, Bruckmaier RM. The effect ofchronic oxytocin-treatment on the bovine mammary gland immunesystem. Vet Med. 2007;52:475–86.

14. Hogan JS, Pankey JW, Duthie AH. Growth inhibition of mastitis pathogensby long-chain fatty acids. J Dairy Sci. 1987;70:927–34.

15. Fanning NF, Porter J, Shorten GD, Kirwan WO, Bouchier-Hayes D, Cotter TG,Phil D, Redmond HP. Inhibition of neutrophil apoptosis after electivesurgery. Surgery. 1999;126(3):527–34.

16. Oshima K, Yamauchi T, Shimomura M. Purification, primary structures, andantibacterial activities of beta-defensins, a new family of antimicrobialpeptides from bovine neutrophils. J Biol Chem. 1996;271:16430.

17. Pinotti L, Rebucci R, Fusi E. Milk choline, alpha-tocopherol andneutrophil chemotaxis in the periparturient dairy cow. Vet ResCommun. 2003;27:265–8.

18. Sreeramkumar V, Adrover JM, Ballesteros I. Neutrophils scan for activatedplatelets to initiate inflammation. Science. 2014;346:1234–8.

Guan et al. BMC Veterinary Research (2020) 16:136 Page 10 of 11

19. Park YH, Fox LK, Hamilton MJ. Suppression of proliferative response ofBoCD4+ T lymphocytes by activated BoCD8+ T lymphocytes in themammary gland of cows with Staphylococcus aureus mastitis. Vet ImmunolImmunopathol. 1993;36:137.

20. Li Y, Ding HY, Wang XC. An association between the level of oxidativestress and the concentrations of NEFA and BHBA in the plasma of ketoticdairy cows. J Anim Physion Anim Nutr. 2016;100:844–51.

21. Zhou Z, Loor JJ, Piccioli-Cappelli F. Circulating amino acids during theperiportal period in cows with different liver functionality index. J Dairy Sci.2016;99:2257–67.

22. Youssef M, El-Ashker M. Significance of insulin resistance and oxidativestress in dairy cattle with subclinical ketosis during the transition period.Trop Anim Health Prod. 2016;49:1–6.

23. Nogalski Z, Wroński M, Lewandowska B. Changes in the blood indicatorsand body condition of high yielding Holstein cows with retained placentaand ketosis. Acta Vet Brno. 2013;81:359–64.

24. Moyes KM, Drackley JK, Salak-Johnson JL, Morin DE, Hope JC, Loor JJ.Dietary-induced negative energy balance has minimal effects on innateimmunity during a streptococcus uberis mastitis challenge in dairy cowsduring mid-lactation. J Dairy Sci. 2009;92:4301–16.

25. Bae H, Jeong CH, Cheng WN, Hong K, Seo HG, Han SG. Oxidative stress-induced inflammatory responses and effects of n-acetylcysteine in bovinemammary alveolar cells. J Dairy Res. 2017;84:418–25.

26. Evangelista V, Piccardoni P, White JG. Cathepsin G-dependent plateletstimulation by activated polymorphonuclear leukocytes and its inhibition byanti-proteinases: role of P-selectin-mediated cell-cell adhesion. Blood. 1993;81:2947.

27. Kishida Y, Kawahara M, Teramukai S, Kubota K, Komuta K, Minato K.Chemotherapy-induced neutropenia as a prognostic factor in advancednon-small-cell lung cancer: results from Japan multinational trialorganization LC00-03. Br J Cancer. 2009;101:1537–42.

28. Li MX, Liu XM, Zhang XF. Prognostic role of neutrophil-to-lymphocyte ratioin colorectal cancer: a systematic review and meta-analysis. Int J Cancer.2014;134:2403–13.

29. Kahramanca S, Ozgehan G, Seker D, Gökce EI, Şeker G, Tunç T, et al.Neutrophil-to-lymphocyte ratio as a predictor of acute appendicitis. UlusTravma Acil Cerr Derg. 2014;20(1):19.

30. Liu WY, Lin SG, Wang LR. Platelet-to-lymphocyte ratio: a novel prognosticfactor for prediction of 90-day outcomes in critically III patients withdiabetic ketoacidosis. Medicine. 2016;95:e2596.

31. Gary T, Pichler M, Belaj K. Platelet-to-lymphocyte ratio: a novel marker forcritical limb ischemia in peripheral arterial occlusive disease patients. PLoSOne. 2013;8:e67688.

32. Sun H, Hu P, Shen H, Dong W, Zhang T, Liu Q, et al. Albumin andneutrophil combined prognostic grade as a new prognostic factor in non-small cell lung cancer: results from a large consecutive cohort. PLoS One.2015;10:e0144663.

33. Koya T, Miyahara N, Takeda K. CD8+ T cell-mediated airwayhyperresponsiveness and inflammation is dependent on CD4+IL-4+ T cells. JImmunol. 2007;179:2787.

34. Suwa T, Hogg JC, Quinlan KB. The effect of interleukin-6 on L-selectin levelson polymorphonuclear leukocytes. Am J Physiol Heart Circ Physiol. 2002;283:879–84.

35. Asensi V, Valle E, Meana A, Fierer J, Celada A, Alvarez V, et al. In vivointerleukin-6 protects neutrophils from apoptosis in osteomyelitis. InfectImmun. 2004;72:3823–8.

36. Ottonello L, Frumento G, Arduino N. Differential regulation of spontaneousand immune complex-induced neutrophil apoptosis by proinflammatorycytokines. Role of oxidants, Bax and caspase-3. J Leukocyte Biol. 2002;72:125.

37. Li L. Signal transduction of neutrophil respiratory burst induced by TNF-α.Int J Immunol. 2000;3:157–60.

38. Lee SW, Hong YS, Chun CM. Anti-inflammatory effects of IL-4 and IL-10 onhuman polymorphonuclear leukocytes. J Korean Med Sci. 2002;17:7–14.

39. Koziol-Montewka M, Ksia̧z̀ek A. Balance between proinflammatory IL1, TNFα,IL6 and regulatory TGFβ, IL10 cytokine serum levels in chronic hemodialysispatients. Immunol Lett. 1997;56:413.

40. National Research Council. Nutrient Requirements of Dairy Cattle. 7th ed.Washington: Nat. Acad. Press; 2001.

41. Greiner M, Pfeiffer D, Smith RD. Principles and practical application of thereceiver-operating characteristic analysis for diagnostic tests. Prevent VetMed. 2000;45:23–41.

Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.

Guan et al. BMC Veterinary Research (2020) 16:136 Page 11 of 11